Self-Organizing Maps for Cellular In Silico Staining and Cell Substate Classification

Neural-net-based cell deconvolution from DNA methylation reveals tumor microenvironment associated with cancer prognosis

DNA methylation is a pivotal epigenetic modification that defines cellular identity. While cell deconvolution utilizing this information is considered useful for clinical practice, current methods for deconvolution are limited in their accuracy and resolution. In this study, we collected DNA methylation data from 945 human samples derived from various tissues and tumor-infiltrating immune cells and trained a neural network model with them. The model, termed MEnet, predicted abundance of cell population together with the detailed immune cell status from bulk DNA methylation data, and showed consistency to those of flow cytometry and histochemistry. MEnet was superior to the existing methods in the accuracy, speed, and detectable cell diversity, and could be applicable for peripheral blood, tumors, cell-free DNA, and formalin-fixed paraffin-embedded sections. Furthermore, by applying MEnet to 72 intrahepatic cholangiocarcinoma samples, we identified immune cell profiles associated with cancer prognosis. We believe that cell deconvolution by MEnet has the potential for use in clinical settings.

Expanding the coverage of spatial proteomics: a machine learning approach

MOTIVATION

Multiplexed protein imaging methods use a chosen set of markers and provide valuable information about complex tissue structure and cellular heterogeneity. However, the number of markers that can be measured in the same tissue sample is inherently limited.

RESULTS

In this paper, we present an efficient method to choose a minimal predictive subset of markers that for the first time allows the prediction of full images for a much larger set of markers. We demonstrate that our approach also outperforms previous methods for predicting cell-level protein composition. Most importantly, we demonstrate that our approach can be used to select a marker set that enables prediction of a much larger set than could be measured concurrently.

AVAILABILITY AND IMPLEMENTATION

All code and intermediate results are available in a Reproducible Research Archive at https://github.com/murphygroup/CODEXPanelOptimization.

Characterization and digital spatial deconvolution of the immune microenvironment of intraductal oncocytic papillary neoplasms (IOPN) of the pancreas

Intraductal oncocytic papillary neoplasm (IOPN) of the pancreas is a distinct entity from intraductal papillary mucinous neoplasms (IPMNs) and is considered one of the precursor lesions of pancreatic cancer. Through immunohistochemistry (IHC) and an artificial intelligence (AI)-based approach, this study aims at characterizing its immune microenvironment. Whole-slide IHC was performed on a cohort of 15 IOPNs, 2 of which harboring an associated adenocarcinoma. The following markers were tested: CD3, CD4, CD8, CD20, CD68, CD163, PD-1, PD-L1, MLH1, PMS2, MSH2, and MSH6. The main findings can be summarized as follows: (i) CD8+ T lymphocytes were the predominant immune cells (p < 0.01); (ii) the vast majority of macrophages were concurrently CD68+ and CD163+; (iii) all tumors showed an activated PD-1/PD-L1 axis, but none had mismatch repair deficiency; (iv) AI-based analysis revealed the presence of 2 distinct regions in each case, namely, Re1, localized at the center of the tumor, and Re2, located at tumor periphery; (v) the infiltrating component of the 2 invasive IOPNs showed a smaller extent of Re1 and a reduced rate of CD4+ cells, as well as a larger extent of Re2 and increased rate of CD8+ cells. IOPNs are lesions enriched in immune cells, with a predominance of CD8+ T lymphocytes and class 2 macrophages. Differently from IPMN-oncogenesis, the progression towards invasive carcinoma is accompanied by an increased rate of CD8+ lymphocytes. This finding may suggest the presence of an active self-immune surveillance in invasive IOPNs, potentially explaining, at least in part, the excellent survival rate of IOPN patients.